CHANGES.html
author Mathieu Lacage <mathieu.lacage@sophia.inria.fr>
Mon, 11 Aug 2008 10:40:31 -0700
changeset 3525 4b8ef7da0ac3
parent 3508 7a3c6148c22d
child 3551 fbf628eb3ca6
permissions -rw-r--r--
bug 275: Queue::Peek should return Ptr<const Packet>

<!doctype html public "-//w3c//dtd html 4.0 transitional//en">
<html>
<head>
   <meta http-equiv="Content-Type" content="text/html; charset=iso-8859-1">
   <title>ns-3 Change Log</title>
</head>
<body>

<h1>
ns-3: API and model change history</h1>
<!--
This ChangeLog is updated in the reverse order 
with the most recent changes coming first.  Date format:  DD-MM-YYYY
-->
<p>
ns-3 is an evolving system and there will be API or behavioral changes
from time to time.   Users who try to use scripts or models across
versions of ns-3 may encounter problems at compile time, run time, or
may see the simulation output change.  </p>
<p>
We have adopted the development policy that we are going to try to ease
the impact of these changes on users by documenting these changes in a
single place (this file), and not by providing a temporary or permanent
backward-compatibility software layer.  </p>
<p>
The goal is that users who encounter a problem when trying to use older
code with newer code should be able to consult this file to find
guidance as to how to fix the problem.  For instance, if a method name
or signature has changed, it should be stated what the new replacement
name is. </p>
<p>
Note that users who upgrade the simulator across versions, or who work
directly out of the development tree, may find that simulation output
changes even when the compilation doesn't break, such as when a
simulator default value is changed.  Therefore, it is good practice for
_anyone_ using code across multiple ns-3 releases to consult this file,
as well as the RELEASE_NOTES, to understand what has changed over time.
</p>
<p>
This file is a best-effort approach to solving this issue; we will do
our best but can guarantee that there will be things that fall through
the cracks, unfortunately.  If you, as a user, can suggest improvements
to this file based on your experience, please contribute a patch or drop
us a note on ns-developers mailing list.  </p>

<hr>
<h1>changes from ns-3.1 to ns-3.2</h1>
<h2>new API in existing classes:</h2>

<ul>
<li>01-08-2008; changeset 
<a href="http://code.nsnam.org/ns-3-dev/rev/a18520551cdf">a18520551cdf</a></li>
<ul>
<li>class ArpCache has two new attributes:  MaxRetries 
and a Drop trace.  It also has some new public methods but these are 
mostly for internal use.
</ul>
</li>
</ul>

<h2>changes to existing API:</h2>

<ul>
<li>04-08-2008; changeset 
<a href="http://code.nsnam.org/ns-3-dev/rev/cba7b2b80fe8">cba7b2b80fe8</a></li>
<ul>
<li>
Cleanup of MTU confusion and initialization in CsmaNetDevice
<pre>
The MTU of the CsmaNetDevice defaulted to 65535.  This did not correspond with
the expected MTU found in Ethernet-like devices.  Also there was not clear 
documentation regarding which MTU was being set.  There are two MTU here, one
at the MAC level and one at the PHY level.  We split out the MTU setting to make
this more clear and set the default PHY level MTU to 1500 to be more like
Ethernet.  The encapsulation mode defaults to LLC/SNAP which then puts the
MAC level MTU at 1492 by default.  We allow users to now set the encapsulation
mode, MAC MTU and PHY MTU while keeping the three values consistent.  See the
Doxygen of CsmaNetDevice::SetMaxPayloadLength for a detailed description of the
issues and solution.
</pre>
</li>
</ul>
</li>
</ul>

<ul>
<li>21-07-2008; changeset 
<a href="
http://code.nsnam.org/ns-3-dev/rev/99698bc858e8">99698bc858e8</a></li>
<ul>
<li> class NetDevice has added a pure virtual method that
must be implemented by all subclasses:
<pre>
virtual void SetPromiscReceiveCallback (PromiscReceiveCallback cb) = 0;
</pre>
All NetDevices must support this method, and must call this callback
when processing packets in the receive direction (the appropriate place
to call this is device-dependent).  An approach to stub this out
temporarily, if you do not care about immediately enabling this
functionality, would be to add this to your device:
<pre>
void
ExampleNetDevice::SetPromiscReceiveCallback
(NetDevice::PromiscReceiveCallback cb)
{ 
  NS_ASSERT_MSG (false, "No implementation yet for
SetPromiscReceiveCallback");
}
</pre>
To implement this properly, consult the CsmaNetDevice for examples of
when the m_promiscRxCallback is called.
</li>
</ul>
</li>
</ul>

<ul>
<li>03-07-2008; changeset 
<a href="http://code.nsnam.org/ns-3-dev/rev/d5f8e5fae1c6">d5f8e5fae1c6</a></li>
<ul>
<li>
Miscellaneous cleanup of Udp Helper API, to fix bug 234
<pre>
class UdpEchoServerHelper
{
public:
- UdpEchoServerHelper ();
- void SetPort (uint16_t port); 
+ UdpEchoServerHelper (uint16_t port);
+ 
+ void SetAttribute (std::string name, const AttributeValue &value);
ApplicationContainer Install (NodeContainer c);

class UdpEchoClientHelper
{
public:
- UdpEchoClientHelper ();
+ UdpEchoClientHelper (Ipv4Address ip, uint16_t port);
- void SetRemote (Ipv4Address ip, uint16_t port);
- void SetAppAttribute (std::string name, const AttributeValue &value);
+ void SetAttribute (std::string name, const AttributeValue &value);
ApplicationContainer Install (NodeContainer c);
</pre>
</li>
</ul>
</li>
</ul>

<ul>
<li>03-07-2008; changeset 
<a href="
http://code.nsnam.org/ns-3-dev/rev/3cdd9d60f7c7">3cdd9d60f7c7</a></li>
<ul>
<li>
Rename all instances method names using "Set..Parameter" to "Set..Attribute"
(bug 232)
</li>
<li> How to fix your code:  Any use of helper API that was using a method
"Set...Parameter()" should be changed to read "Set...Attribute()".  e.g.
<pre>
- csma.SetChannelParameter ("DataRate", DataRateValue (5000000));
- csma.SetChannelParameter ("Delay", TimeValue (MilliSeconds (2)));
+ csma.SetChannelAttribute ("DataRate", DataRateValue (5000000));
+ csma.SetChannelAttribute ("Delay", TimeValue (MilliSeconds (2)));
</pre>
</li>
</ul>
</li>
</ul>

<h2>changed behavior:</h2>
<ul>
<li> 28-07-2008; changeset 
<a href="http://code.nsnam.org/ns-3-dev/rev/6f68f1044df1">6f68f1044df1</a>
<ul>
<li>
OLSR: HELLO messages hold time changed to 3*hello
interval from hello interval.  This is an important bug fix as
hold time == refresh time was never intentional, as it leads to
instability in neighbor detection.
</ul>
</li>
</ul>
</li>

</body>
</html>